Although methylglyoxal is derived from glycolysis, it has adverse effects on cellular function. Hence, the intrinsic role of methylglyoxal in vivo remains to be determined. Glyoxalase 1 is a pivotal enzyme in the metabolism of methylglyoxal in all types of organisms. To learn about the physiological roles of methylglyoxal, we have screened conditions that alter the expression of the gene encoding glyoxalase 1, GLO1, in Saccharomyces cerevisiae. We show that the expression of GLO1 is induced following treatment with Ca2+ and is dependent on the MAPK (mitogen-activated protein kinase) Hog1 protein and the Msn2/Msn4 transcription factors. Intriguingly, the Ca2+-induced expression of GLO1 was enhanced in the presence of FK506, a potent inhibitor of calcineurin. Consequently, the Ca2+-induced expression of GLO1 in a mutant that is defective in calcineurin or Crz1, the sole transcription factor downstream of calcineurin, was much greater than that in the wild-type strain even without FK506. This phenomenon was dependent upon a cis-element, the STRE (stress-response element), in the promoter that is able to mediate the response to Ca2+ signalling together with Hog1 and Msn2/Msn4. The level of Ca2+-induced expression of GLO1 reached a maximum in cells overexpressing MSN2 even when FK506 was not present, whereas in cells overexpressing CRZ1 the level was greatly reduced and increased markedly when FK506 was present. We also found that the levels of Msn2 and Msn4 proteins in Ca2+-treated cells decreased gradually and that FK506 blocked the degradation of Msn2/Msn4. We propose that Crz1 destabilizes Msn2/Msn4 in the nuclei of cells in response to Ca2+ signalling.
MG (methylglyoxal) is a typical 2-oxoaldehyde that is synthesized during glycolysis, a ubiquitous energy-generating pathway [1,2]. Although MG is a natural metabolite it can inhibit the growth of cells [1–3] and in some cases induce apoptosis or necrosis, with the mode of cell death depending upon the cell line examined [4–6]. MG was once believed to be a major intermediate of glycolysis, but that hypothesis has been disproved; nevertheless, MG and its metabolic enzymes have received considerable attention, because MG is involved in diseases such as diabetes and its complications, cancers, Alzheimer's disease and autism [7–12]. Hence, there must be an ingenious system to avoid any metabolic disorder involving MG in cells, thereby reducing the risk of suffering such diseases. However, the molecular mechanisms by which MG causes such diseases are not well understood.
To find a clue as to the physiological role of MG, we searched for conditions that alter intracellular levels of MG, as well as the activities of MG-metabolizing enzymes in the yeast Saccharomyces cerevisiae. We have reported previously that the expression of GLO1 is specifically induced by osmotic stress via the MAPK (mitogen-activated protein kinase) Hog1  and that expression of GLO1 is crucial for the metabolism of MG . The GLO1 promoter contains a characteristic cis-acting element, the STRE (stress-response element). Genes possessing the STRE usually respond to a wide variety of stress stimuli, such as oxidative stress, heat shock stress and osmotic stress . MSN2 and MSN4, originally cloned as multicopy suppressors of the SNF1 mutation, encode C2H2-type zinc finger transcription factors that have a high degree of functional redundancy; they have been shown to translocate into the nucleus, and bind to the STRE under environmental stress [15a]. Even though the GLO1 promoter possesses two STREs, intriguingly, GLO1 did not respond to any stress other than osmotic stress . S. cerevisiae produces glycerol as a compatible osmolyte under high-osmotic stress . We found that the uptake of glucose, and subsequently, the flux of glucose into glycolysis, was enhanced when cells were exposed to high-osmotic stress in order to facilitate glycerol production . Consequently, intracellular MG increased , because the major source of MG is glycolysis. Therefore we have proposed previously that the physiological significance of the specific expression of GLO1 under high-osmotic stress is probably due to the efficient metabolism of MG, which increases during the response to osmotic stress [2,13].
On the other hand, we have studied the effect of MG on cellular function. As a result, we revealed that MG functions as a signal initiator in yeasts [17–19]. For example, we found that the AP-1 (activator protein 1)-like bZIP (basic leucine zipper) transcription factor Yap1 is constitutively activated in glo1Δ cells. This occurred via MG-mediated modification of cysteine residues and these residues are crucial for the determination of its nucleocytoplasmic localization . Yap1 also plays crucial roles in the response to oxidative stress, as well as drug stress in S. cerevisiae (for a review see ). Recently, we found that MG attenuates overall protein synthesis through phosphorylation of the translation initiation factor eIF2α in a TOR (target of rapamycin)-independent manner , and another AP-1-like transcription factor Gcn4 plays a role in the adaptive response to MG stress . These results show that cellular levels of MG must be controlled adequately to warrant normal cellular functions.
In the present study we show that Ca2+ induces the expression of GLO1, and that the induction is strictly dependent upon Hog1, Msn2/Msn4 and the two STREs in the GLO1 promoter. Notably, the Ca2+-induced expression of GLO1 was further augmented by FK506, a potent inhibitor of calcineurin . Calcineurin is a protein phosphatase whose activity is regulated by Ca2+/calmodulin. Full induction of the expression of GLO1 following treatment with Ca2+ was observed in a mutant defective in either calcineurin or Crz1, which encodes the sole transcription factor functioning under the control of calcineurin [24–27]. We also provide evidence that Crz1 negatively affects the functions of Msn2/Msn4 in the nuclei of cells treated with Ca2+ by facilitating the degradation of Msn2 and Msn4. Consequently, a further increase in the Ca2+-induced expression of GLO1 was seen when either FK506 was present, or calcineurin or Crz1 was disrupted. Our results suggest that Crz1 destabilizes Msn2 and Msn4 proteins in the nucleus when cells are treated with Ca2+.
Strain, plasmids and medium
Unless otherwise stated, yeast strains used in the present study have the YPH250 background (MATa trp1-Δ1 his3-Δ200 leu2-Δ1 lys2–801 ade2–101 ura3–52). Construction of the hog1Δ, pbs2Δ, ssk1Δ, sho1Δ, ssk1Δsho1Δ, cnb1Δ, crz1Δ, msn2Δ and msn2Δmsn4Δ mutants in YPH250 has been described previously [13,18]. The erg6Δ (deficient in ergosterol biosynthesis) mutant has the BY4741 background (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0). Cells were cultured in YPD medium [1% (w/v) yeast extract/2% (w/v) peptone/2% (w/v) glucose] or SD (synthetic dextrose) minimal medium [2% (w/v) glucose and 0.67% yeast nitrogen base without amino acids] with appropriate amino acids and bases at 28 °C with reciprocal shaking.
The preparation of cell extracts and measurement of the activities of Glo1 and β-galactosidase were as described previously [13,14]. One unit of the activity of Glo1 was defined as the amount of enzyme forming 1 μmol of S-D-lactoylglutathione per min using a millimolar absorption coefficient of 3.37 mM−1·cm−1 . One unit of β-galactosidase activity was defined as the amount of enzyme that increases the A420 by 1000 per h . Protein concentrations were determined by the Bradford method .
Cells were cultured in YPD medium until the D610 was 1.0, and 300 mM CaCl2 and/or 1 μg/ml FK506 were added. After 30 min of incubation, total RNA was prepared according to the method described by Schmitt et al. . The DNA probe, labelled with [α-32P]dCTP was prepared as described previously .
Detection of Hog1 phosphorylation
Cells were cultured in YPD medium until the D610 was 1.0, and 300 mM CaCl2 and/or 1 μg/ml FK506 were added. After 60 min of incubation, cell lysates were prepared as described by Bell et al. . Yeast cellular proteins were separated by SDS/PAGE (10% gels) and transferred on to a PVDF membrane. Anti-phospho-p38 monoclonal antibody (Sigma–Aldrich) was used as the primary antibody and anti-(rabbit IgG) antibody conjugated with horseradish peroxidase (New England Biolabs) was used as the secondary antibody. To measure the level of Hog1 protein, an anti-Hog1 antibody (Santa Cruz Biotechnology) was used as a primary antibody.
Detection of Msn2 and Msn4
To determine the levels of Msn2 and Msn4 proteins, cells carrying pRS316+Msn2-GFP or pGR247 (pAdh1-Msn4-GFP)  were cultured in YPD medium until they were growing exponentially and they were then treated with 300 mM CaCl2. Cells were collected (2000 g centrifugation for 15 s) at the indicated prescribed time, and disrupted with glass beads in 100 mM Tris/HCl buffer, pH 7.0, containing 5 mM MgCl2, 10% (v/v) glycerol, 0.1% Triton X-100, 50 mM NaCl, 1 mM dithiothreitol and protease inhibitor cocktail (Nacalai Tesque). Cell extracts (20 μg of protein) were subjected to SDS/PAGE (7.5% gels). Anti-GFP (green fluorescent protein) antibody (Santa Cruz Biotechnology) was used to detect Msn2–GFP and Msn4–GFP. Detection of Cdc28 was used as a loading control [the membrane was treated in 0.2 M glycine (pH 2) for 5 min and reprobed with anti-Cdc2 antibody (Santa Cruz Biotechnology)].
Construction of GLO1–lacZ reporter gene
The GLO1–lacZ cassette in pRSGlac415  was cloned into the SalI site of YIp5 (to give YIp5+GLO1-lacZ), and the cassette was then amplified by PCR with primers 5TO358F (5′-AGGGCATCGGTCGACGGATCCGGGTAATTC-3′) and 5TO358R (5′-TAAAACGACGGAATTCCCGGGTTTCTCAAT-3′) using pYIp5+GLO1-lacZ as a template. SalI and EcoRI sites were added to 5TO358F and 5TO358R respectively (underlined nucleotides). The PCR fragment was digested with SalI and EcoRI, and cloned into the SalI/EcoRI site of YIp358R. The resultant plasmid (YIp358R+GLO1-lacZ) was digested with NcoI and the linearized fragment was integrated into the ura3 locus of YPH250.
To insert a point mutation into the STREs (5′-AGGGG-3′→5′-AGATG-3′), an overlap extension PCR was conducted . For the site-directed mutagenesis, four internal primers (GSTRE1, 5′-AATAGGTAAAGAGATGGGTGGGGGTGG-3′; GSTRE1R, 5′-CCACCCCCACCCATCTCTTTACCTATT-3′; GSTRE2, 5′-CTGAATAAACAAGATGCTTTACGATGG-3′; and GSTRE2R, 5′-CCATCGTAAAGCATCTTGTTTATTCAG-3′) were designed. To create the mutation in STRE1 (from −432 to −428 relative to the initiation ATG codon), the first PCR was performed with 5TO358F plus GSTRE1R, and 5TO358R plus GSTRE1. The second PCR was performed with 5TO358F and 5TO358R as the primers and a mixture of the products of each first PCR as the template. The PCR product was digested with SalI and EcoRI and the resultant fragment was inserted into the SalI/EcoRI site of YIp358R to yield YIp358R+PMS1. Similarly, to create the mutation in STRE2 (from −229 to −225), the first PCR was performed with 5TO358F plus GSTRE2R and 5TO358R plus GSTRE2. The second PCR was performed with 5TO358F and 5TO358R and a mixture of the products of each first PCR as the template. The PCR product was cloned into YIp358R as described above (YIp358R+PMS2). To construct the GLO1–lacZ reporter gene carrying both mutated STRE1 and STRE2 (PMS12), the first PCR was performed with 5TO358F plus GSTRE2R and 5TO358R plus GSTRE2 using YIp358R+PMS1 as a template. The second PCR was performed using 5TO358F and 5TO358R and a mixture of the products of each first PCR as the template. The PCR product was digested with SalI and EcoRI, and the resultant fragment was inserted into the SalI/EcoRI site of YIp358R to yield YIp358R+PMS12. The correct introduction of the mutation was verified by DNA sequencing.
Construction of glutathione peroxidase reporter gene GPX1–lacZ
To amplify the region between −365 and +8 of GPX1 (relative to the translation initiation ATG codon), PCR was performed with primer GPX1–lacZ-F (5′-GGAGTCGACGGACTTGATAGAATCCACCTT-3′) and GPX1-lacZ-R (5′- GGTGAAAAAGAATGAATTCCTTGCATCGTT-3′) using the genomic DNA of YPH250 as a template. SalI and EcoRI sites were designed in GPX1–lacZ-F and GPX1-lacZ-R respectively (underlined nucleotides). The PCR fragment was digested with SalI and EcoRI, and cloned into the SalI/EcoRI site of YIp358R. The resultant plasmid (YIp358R-GPX1-lacZ) was digested with NcoI, and the linearized fragment was integrated into the ura3 locus of YPH250.
Construction of HOG1-overexpression plasmid
The HOG1 gene was amplified by PCR with primers HOG1S (5′-GTTGTTAGGAAAGCATGCTTTATCTCCAAG-3′) and HOG1R (5′-CCTTTTATGGGATCCTAATTTCTTAAGGAG-3′) using the genomic DNA of YPH250 as a template. SphI and BamHI sites were designed in HOG1S and HOG1R respectively (underlined nucleotides). The PCR fragment was digested with SphI and BamHI, and cloned into the SphI/BamHI sites of plasmid pRS423 to yield pRS423+HOG1.
Cells expressing GFP-tagged proteins were cultured in YPD medium to a D610 of 0.5, and 300 mM CaCl2 was added in the presence or absence of 1 μg/ml FK506. After incubation for the prescribed time at 28 °C, the intracellular localization of each GFP-tagged protein was observed using a fluorescence microscope (Olympus BX51). Plasmids carrying Hog1–GFP, Msn2–GFP and Crz1–GFP were pRS4-Hog1-GFP , pAMG (pAdh1-Msn2-GFP)  and pASM463  respectively.
Expression of GLO1 is induced by Ca2+ in a Hog1 MAPK cascade-dependent manner
We first searched for conditions under which either the cellular concentration of MG is changed or the expression of GLO1, which encodes an enzyme crucial for the metabolism of MG [2,13], is altered. We found that levels of GLO1 mRNA were increased following treatment with CaCl2 (Figure 1A). We therefore constructed a GLO1–lacZ reporter gene to verify the Ca2+-induced expression of GLO1 and found that the activity of β-galactosidase driven by GLO1–lacZ was increased following treatment with CaCl2 (see Figure 5B). The activity of Glo1 was also increased in cells treated with CaCl2 (Figure 1B). Therefore we concluded that Ca2+ induces the expression of GLO1.
Hog1 is one of the MAPKs in S. cerevisiae and plays a crucial role in the response to osmotic stress (for reviews see [35,36]). Therefore we conducted a Northern blot analysis to determine the GLO1 mRNA level in a hog1Δ mutant. As shown in Figure 1(A), the level of GLO1 mRNA was not increased in hog1Δ cells following the treatment with CaCl2. Induction of Glo1 activity by CaCl2 also did not occur in the hog1Δ nor the pbs2Δ mutant, which is defective in a MAPK kinase responsible for the activation of Hog1 (Figure 1B).
Besides Hog1, we have reported previously that the Msn2 and Msn4 C2H2 zinc finger transcription factors are necessary for the up-regulation of GLO1 in response to osmotic stress . To determine whether these transcription factors are also involved in the Ca2+-induced expression of GLO1, we conducted Northern blotting in an msn2Δmsn4Δ double mutant. As shown in Figure 1(A), the levels of GLO1 mRNA in msn2Δmsn4Δ cells did not increase following the treatment with CaCl2, indicating that Msn2 and Msn4 are necessary for the response of GLO1 to Ca2+.
Next, we determined the roles of the osmosensors functioning upstream of the Hog1 MAPK cascade. In S. cerevisiae, two such osmosensors, Sln1 and Sho1, have been thus far identified. The Sln1 branch consists of the phospho-relay protein Ypd1 and the response regulator Ssk1, constituting a two-component system . Two redundant MAPK kinase kinases (Ssk2 and Ssk22) function downstream of the Sln1 branch. In the Sho1 branch, Sho1 physically interacts with Pbs2 through the SH3 (Src homology 3) domain [38,39]. To determine whether these two osmosensors are involved in the Ca2+-induced expression of GLO1, we measured the Glo1 activity in cells defective in these branches. As SLN1 is an essential gene, we disrupted SSK1 and/or SHO1 to inactivate the osmotic-stress signalling pathways upstream of Hog1. As shown in Figure 1(B), the induction of GLO1 expression was observed in mutants defective in either SSK1 or SHO1, although it was repressed in the ssk1Δsho1Δ double mutant.
Response of GLO1 to Ca2+ and FK506
Although we have shown that the Hog1 MAPK cascade is necessary for the Ca2+-induced expression of GLO1, we also set out to determine whether the Ca2+/calmodulin-dependent calcineurin system was also involved in this response, because the system is a well-known pathway mediating Ca2+ signalling in eukaryotic cells, including in S. cerevisiae . In the Ca2+/calcineurin system, extracellular Ca2+ enters the cell and binds calmodulin to activate a protein phosphatase calcineurin, which in turn dephosphorylates Crz1, a transcription factor functioning downstream of calcineurin [24,25]. As FK506 is a potent inhibitor of calcineurin , the expression of the Ca2+-induced genes regulated by the Ca2+/calcineurin system, e.g. GSC2 (glucan synthase of cerevisiae, also known as FKS2) and GPX2 [41,42] is usually markedly repressed in its presence. To address whether the calcineurin system is involved in the Ca2+-induced expression of GLO1, we determined the effect of FK506. Surprisingly, and rather bizarrely, the Ca2+-induced expression of GLO1 mRNA was further increased when FK506 was present (Figure 1A). Similarly, the activities of β-galactosidase, driven by the GLO1–lacZ reporter, as well as Glo1 were further increased by the simultaneous treatment of cells with CaCl2 and FK506 (Figures 1B and 5B). This was observed in the ssk1Δ and sho1Δ mutants, but not in the ssk1Δsho1Δ mutant (Figure 1B). These results show that two osmosensors function as the upstream modules for the transduction of the Ca2+ signal to Hog1, and subsequently, that there was a bizarre response, whereby Glo1 is induced by Ca2+ and FK506.
Calcineurin/Crz1 pathway is involved in the bizarre response of GLO1
We confirmed that the bizarre response of GLO1 to Ca2+ and FK506 as described above was also seen when employing much lower concentrations (50 mM) of CaCl2 (results not shown). However, as the maximal induction of GLO1 is brought about by 200–300 mM CaCl2, we subsequently analysed this response in the presence of 300 mM CaCl2.
To verify whether the bizarre response caused by FK506 in terms of the Ca2+-induced expression of GLO1 was brought about by a certain action of FK506 on the machinery involved in Ca2+ signalling or whether it was dependent upon the calcineurin system we conducted a genetic analysis. We disrupted both the regulatory subunit of calcineurin (CNB1), to abolish the phosphatase activity , and also Crz1. As shown in Figure 2(A), the expression of GLO1 in both cnb1Δ and crz1Δ mutants reached a maximum level on treatment with CaCl2 alone, and the treatment of these mutants with CaCl2 and FK506 simultaneously did not enhance the expression of GLO1 any further. Thereby we confirmed that the effect of FK506 on the expression of GLO1 with respect to the response to Ca2+ was exerted through the calcineurin/Crz1 pathway.
This positive effect of FK506 was not observed for the NaCl-induced expression of Glo1 (Figure 2B). This was also the case for the KCl- and sorbitol-induced expression of GLO1 mRNA (results not shown). Furthermore, basal levels of GLO1 mRNA, as well as Glo1 activity, did not increase in the cnb1Δ and crz1Δ mutants (Figures 2A and 2B), therefore the calcineurin/Crz1 pathway does not act as a negative regulator of the expression of GLO1 under normal conditions. It is likely that the effect of FK506 on the Hog1-dependent induction of the expression of GLO1 occurs only when Ca2+ is present, and FK506 does not enhance the response to osmotic stress in general. Therefore we temporally referred to this phenomenon as the ‘bizarre response’, because Crz1 is a transcription factor that essentially has a positive effect on its target gene, but nevertheless, Crz1 acted as a negative regulator for the Ca2+-induced expression of GLO1.
One possible explanation for the bizarre response is that some newly synthesized protein(s), the synthesis of which is regulated by the calcineurin/Crz1 pathway, function as a negative regulator of the expression of GLO1 in response to Ca2+. To address this possibility, we treated yeast cells with CaCl2 and FK506 in the presence of cycloheximide. As shown in Figure 2(C), the bizarre response was observed even under conditions where protein synthesis is blocked. Hence, the bizarre response is likely to be caused by the pre-existing machinery in yeast cells.
Time course of the expression of GLO1
To gain further insights into the role of the calcineurin pathway in the Ca2+-induced expression of GLO1, we monitored the level of GLO1 mRNA in cells treated with CaCl2. As shown in Figure 3, the amount of GLO1 mRNA in wild-type cells gradually increased upon the addition of CaCl2, reaching a maximum after 30–40 min of incubation. The timing of the increase in the level of GLO1 mRNA in cells treated with CaCl2 plus FK506 was almost the same as that in cells treated with CaCl2 alone, although the amount of GLO1 mRNA was significantly higher. The pattern of the changes in the level of GLO1 mRNA in cnb1Δ cells treated with CaCl2 was substantially the same as that in wild-type cells treated with CaCl2 plus FK506. The expression profile of GLO1 in crz1Δ cells in response to Ca2+ was essentially similar to that in cnb1Δ cells (results not shown). Therefore the calcineurin pathway seems to negatively influence the maximal level of GLO1 mRNA in response to Ca2+.
Phosphorylation and nucleocytoplasmic localization of Hog1
Hog1 is phosphorylated by Pbs2, and then translocates into the nucleus in response to high osmotic stress [33,44,45]. As the Ca2+-induced expression of GLO1 was strictly dependent upon Hog1, we determined the level of Hog1 phosphorylation following treatment with CaCl2. As shown in Figure 4(A), Hog1 was phosphorylated in cells treated with CaCl2; however, the simultaneous treatment of cells with CaCl2 and FK506 did not enhance the phosphorylation of Hog1. Similarly, the level of phosphorylated Hog1 in cnb1Δ, as well as crz1Δ cells treated with CaCl2, was the same as that in wild-type cells and the addition of FK506 did not give rise to any further increase in Hog1 phosphorylation.
Next, we determined the nucleocytoplasmic localization of Hog1. As shown in Figure 4(B), Hog1 was concentrated in the nucleus upon CaCl2 treatment and the intensity of the fluorescence derived from Hog1–GFP in the nuclei in cells treated with CaCl2 was the same as that observed in cells treated with CaCl2 and FK506 simultaneously.
We also determined the time course of the nuclear localization of Hog1 following the treatment with CaCl2 and FK506 (Figure 4C). In wild-type cells, Hog1 was seen to immediately accumulate in the nucleus upon the treatment with CaCl2 and redistributed to the cytoplasm after 30–45 min of incubation in the presence of CaCl2. No distinct difference was observed in the timing when FK506 was present (Figure 4C). This was also the case in cnb1Δ and crz1Δ cells (Figure 4C). Therefore although Hog1 is necessary for GLO1 to respond to extracellular Ca2+, neither the level of phosphorylation nor the nucleocytoplasmic localization of Hog1 is likely to be a cause of the bizarre response.
Roles of STREs in the GLO1 promoter in the bizarre response
The GLO1 promoter contains two STREs (STRE1 at −432 to −428 and STRE2 at −229 to −225) . As the Ca2+-induced expression of GLO1 was dependent upon the Msn2/Msn4 transcription factors (Figure 1A), and it has been reported that Msn2 and Msn4 bind to the STRE , we next determined whether the STREs are involved in the bizarre response. A point mutation, known to render the STRE non-functional , was introduced into either or both of the STREs (5′-AGGGG-3′→5′-AGATG-3′) (Figure 5A). The Ca2+-induced expression of the GLO1–lacZ reporter was reduced by approx. 45% by introduction of a mutation into either STRE1 or STRE2, and greatly repressed by the simultaneous introduction into both STREs (Figure 5B). Importantly, the bizarre response was repressed by the mutation in either STRE1 or STRE2, although STRE2 seemed to have a more important role in the response.
Next, we determined the response of GLO1–lacZ reporters with various STRE constructs in an msn2Δ mutant. As Msn2 is the primary transcription factor for the STRE-dependent gene, the Ca2+-induced expression of GLO1–lacZ was impaired in msn2Δ cells when compared with wild-type cells (Supplementary Figure S1 available at http://www.BiochemJ.org/bj/427/bj4270275add.htm). Although the induction rate was quite small, the Ca2+-induced expression of GLO1 did occur in msn2Δ cells and this may be attributable to activation by Msn4; however, only a limited increase in β-galactosidase activity was observed in the presence of FK506 (Supplementary Figure S1). As the Ca2+-induced expression of PMS1 was comparable with that of the wild-type STRE construct in msn2Δ cells, Msn4 may bind to STRE2 in response to Ca2+. No induction was observed in msn2Δmsn4Δ cells (Figure 1A).
As we have demonstrated that Hog1 plays a crucial role in the bizarre response, we further determined the effect of Hog1-deficiency on the expression of the GLO1–lacZ reporters with various STRE constructs. As shown in Figure 5(C), as Hog1 is crucial, the level of Ca2+-induced expression of GLO1 was markedly reduced in a hog1Δ mutant (note that the vertical scale of Figure 5B is 50-fold that of Figure 5C) and the bizarre response was not clearly observed in the presence of FK506 together with CaCl2, even in the GLO1–lacZ reporter with wild-type STREs. We also determined the response of GLO1–lacZ in Hog1-overexpressing cells. Intriguingly, as shown in Figure 5(D), the bizarre response was not observed.
STREs regulated by both Mns2/Msn4 and Hog1 exhibit the bizarre response
To verify whether the STRE is sufficient for the bizarre response, we used the STRECTT1–lacZ reporter gene; this contained two STREs from the promoter of the catalase-T-encoding gene CTT1 (5′-TTCAAGGGGATCACCGGTAAGGGGCCAAG-3′; the STREs are underlined) followed by the TATA box of CYC1 . As shown in Figure 6(A), the bizarre response was observed with this reporter gene. As the CTT1 gene encodes a cytoplasmic catalase , we measured the catalase activity. The Ca2+-induced increase in catalase activity was further increased by addition of FK506 (Figure 6B). In addition, as observed in Glo1, catalase activity was markedly increased in cnb1Δ, as well as crz1Δ, cells following treatment with CaCl2 alone, but did not increase any further when FK506 was present (Figure 6B).
Very recently, we found that the expression of GPX1, encoding an S. cerevisiae homologue of mammalian glutathione peroxidase , is induced following treatment with CaCl2 in an Msn2/Msn4-dependent manner (Figure 6C) and that the GPX1 promoter contains two functional STREs ; nevertheless, GPX1 did not exhibit the bizarre response (Figure 6C). We therefore have found that Hog1 is not necessary for the Ca2+-mediated induction of GPX1 , whereas, the Ca2+-induced response for both catalase, as well as STRECTT1–lacZ, is dependent upon Hog1 (Figures 6A and 6B). Collectively, these results suggest that STREs regulated by both Hog1 and Msn2/Msn4 mediate the bizarre response.
Effect of Crz1 on the function of Msn2
The results presented so far imply two possibilities: (i) that Crz1 could compete with Msn2 to bind to the STRE in the GLO1 promoter to reduce the effect of Msn2 on the Ca2+-induced expression of GLO1; or (ii) that Crz1 inhibits the activity of Msn2 towards the STRE. It is known that Crz1 binds to a distinct DNA sequence, the CDRE (calcineurin-dependent response element) , the consensus sequence of which (5′-GAGGCTG-3′) does not resemble the STRE. In addition, the GLO1 promoter does not contain a sequence exactly coinciding with that of the CDRE near the STREs, so the first scenario is not likely. To address the possibility of the latter mechansim, the response of GLO1 to Ca2+ was determined in cells overexpressing MSN2 or CRZ1. As shown in Figure 7(A), the activity of Glo1 in Msn2-overexpressing cells following treatment with CaCl2 was increased to a much greater extent than that in cells carrying the vector alone, and the addition of FK506 did not bring about a further increase in the activity. On the other hand, the induction of Glo1 activity by CaCl2 in CRZ1-overexpressing cells was repressed, but the activity increased to a maximal level when CaCl2 and FK506 were added simultaneously.
We have reported previously that Msn2 is accumulated in the nucleus following treatment with CaCl2 . The time course of Msn2 accumulation is shown in Figure 7(B); Msn2 was found in the nucleus immediately upon the treatment of cells with CaCl2 where it resided for up to 30 min and redistribution in the cytoplasm was observed after 45–60 min of incubation. The behaviour of Msn2 in terms of its nucleocytoplasmic localization, i.e. the timing of the beginning of the nuclear accumulation and the retention period in the nucleus, was not affected by the presence of FK506 (Figure 7B). Similarly, the behaviour of Msn2 in cnb1Δ cells was the same as that in wild-type cells (Figure 7B). The nucleocytoplasmic localization of Msn2 in response to Ca2+ and FK506 in crz1Δ cells was substantially the same as that in wild-type and cnb1Δ cells (results not shown). Furthermore, Crz1 also accumulates in the nucleus upon treatment with CaCl2 , but we found that the accumulation was blocked by FK506 (Figure 7C). The nuclear accumulation of Crz1 did not occur following treatment with NaCl (Figure 7C). Therefore the Ca2+-induced Crz1 nuclear accumulation seems to cause this transcription factor to function as a negative regulator of Msn2 in the nucleus.
Crz1 affects the stability of Msn2 in the nucleus in response to Ca2+
Durchschlag et al.  have reported that Msn2 is degraded in the nucleus. Hence, one feasible explanation for the bizarre response is that the nuclear Crz1 affects the stability of Msn2 in the nucleus when Ca2+ is present. To address this possibility, we determined the level of Msn2 protein following treatment with CaCl2 with or without FK506. As shown in Figure 8(A), the level of Msn2 protein in wild-type cells gradually decreased in the presence of CaCl2, but the reduction was blocked when FK506 was present. This is in contrast with crz1Δ cells, where the reduction in the level of Msn2 protein following the treatment with CaCl2 was repressed even in the absence of FK506 (Figure 8A). Substantially, the same results in terms of the stability of Msn2 were obtained in cnb1Δ cells (results not shown). Additionally, the same results were obtained with respect to the stability of Msn4 (Figure 8B). Taken together, Crz1 seems to destabilize Msn2 in response to Ca2+.
As the bizarre response was observed in genes where the expression is dependent upon Hog1, we determined the stability of Msn2 in hog1Δ cells and Hog1-overexpressing cells. The Ca2+-induced degradation of Msn2 was observed in hog1Δ cells and was blocked by the addition of FK506 (Figure 8C). In contrast, in Hog1-overexpressing cells, the Ca2+-induced degradation of Msn2 was not blocked by the presence of FK506 (Figure 8D); this is consistent with the results presented above whereby further increases in the expression of GLO1–lacZ reporter in the presence of CaCl2 and FK506 (i.e. the bizarre response) were not observed in Hog1-overexpressing cells (Figure 5D).
Negative effect of Crz1 on Msn2
To learn about the physiological role of MG in yeast cells, we searched for conditions that alter the expression of GLO1, and revealed that GLO1 is expressed following treatment with CaCl2. During the course of the present study, intriguingly, we found that the Ca2+-induced expression of GLO1 was further enhanced if FK506 was present, which we referred to as the bizarre response.
In the present study, we found that Crz1 negatively influences the function of Msn2 on the STRE in the GLO1 promoter in cells treated with CaCl2. Therefore the bizarre response occurred only with a combination of Ca2+ and FK506, and not with a combination of FK506 and other ions (Na+ and K+) or sorbitol, at concentrations that provoke the nuclear accumulation of Msn2. This is presumably because such chemicals do not activate the calcineurin system, and therefore Crz1 does not accumulate in the nucleus (Figure 7C) preventing interaction with Msn2. However, although several genes have expression that is regulated by Msn2 and the STRE, we found that only Hog1-dependent genes exhibited the bizarre response (Figure 5B). In other words, a gene whose expression is dependent upon these three factors will exhibit the bizarre response (e.g. CTT1, but not GPX1). As calcineurin is a protein phosphatase, a simple model explaining the role of Hog1 in the bizarre response is that calcineurin dephosphorylates Hog1, thereby preventing the Ca2+ signal from reaching the target gene. However, as shown in Figure 4(A), disruption of CNB1 or CRZ1 did not affect the level of phosphorylation or the nuclear accumulation of Hog1. These results support our conclusion that Crz1 affects the function of Msn2/Msn4 and not Hog1.
There are several explanations as to the mechanism whereby Crz1 acts negatively on the expression of genes regulated by Hog1, Msn2/Msn4 and STREs. For example, Crz1 could bind directly to Msn2/Msn4 to reduce its function as a transcription factor. To address this possibility, we tried to detect any direct interaction between Crz1 and Msn2 by conducting a GST (glutathione transferase) pull-down and co-immunoprecipitation assay, however, we were unable to obtain evidence of physical interactions (results not shown).
Another possibility is that Crz1 influences the machinery involved in the nuclear localization of Msn2. However, as shown in Figure 7(B), the behaviour of Msn2, in terms of its nucleocytoplasmic localization in response to Ca2+, was virtually unaffected by FK506. In addition, Durchschlag et al.  have reported that Msn2 is constitutively concentrated in the nucleus in msn5Δ cells because Msn5 is an exportin for Msn2 [34,53]; nevertheless, the basal level of CTT1 mRNA and the stress response of CTT1 expression in the msn5Δ mutant were normal . This is accounted for by a decrease in the total amount of Msn2 protein in msn5Δ cells (see below). The expression of CTT1 under environmental stress is regulated by Hog1, Msn2/Msn4 and the STRE , and consequently the CTT1 promoter shows the bizarre response (Figures 6A and 6B). Although Msn2 is a positive regulator for the expression of GLO1, the basal activity of Glo1 did not increase in the msn5Δ mutant (results not shown), as observed in the case of CTT1, despite the constitutive nuclear accumulation of Msn2. Therefore the cause of the bizarre response is not likely to be the nuclear retention period of Msn2.
Next, we examined the possibility that Crz1 titrates Hog1 away from Msn2, thereby reducing the activity as transcription factor in the presence of CaCl2. Although a direct physical interaction between Crz1 and Hog1 has not been reported so far, if the nuclear Crz1 titrates Hog1 away from Msn2, further increases in the expression of GLO1–lacZ would be expected to occur in the presence of FK506, which blocks the nuclear accumulation of Crz1; however, this was not the case (Figure 5D). Furthermore, in contrast with in Msn2-overexpressing cells (Figure 7A), full activation of Msn2, in terms of the expression of GLO1–lacZ, was not observed in Hog1-overexpressing cells (Figure 5D). We propose that this is because the proportion of Msn2 that is not bound to STRE in the GLO1 promoter is increased. It has been reported that Hog1 forms a complex with Msn2, although Hog1 in itself is not a transcription factor. Therefore overexpression of Hog1 may increase the proportion of Hog1–Msn2 complex that is not bound to the STRE thereby removing Msn2 away from the STRE target sequence; this reduces the opportunity of Msn2 to function as transcription factor on the promoter. Subsequently, full activation is not observed following treatment with CaCl2. Additionally, if Crz1 was titrated by Hog1, the induction of the expression of Crz1-dependent genes by Ca2+ would be impaired in Hog1-overexpressing cells. So, we determined the response of GPX2 to Ca2+, the expression of which is regulated by the calcineurin/Crz1 pathway . However, GPX2 responded normally to Ca2+ even in Hog1-overexpressing cells (Supplementary Figure S2A available at http://www.BiochemJ.org/bj/427/bj4270275add.htm). Taken together, we concluded that Hog1 was not titrated by Crz1 in the presence of Ca2+.
Genome-wide search for the genes showing the bizarre response
Although a typical Crz1-binding consensus sequence (5′-GAGGCTG -3′) has not been found in the GLO1 promoter, Ruiz et al.  have recently reported that a 5′-GGGGCTG-3′ sequence in the promoter of the hexose transporter HXT2 is functional as a CDRE. A similar sequence (5′-GGGGCTT-3′; −228 to −222) overlaps the STRE2 (5′-AGGGG-3′; −229 to −225) in the GLO1 promoter. So, to address the possibility that the bizarre response is mediated by the co-existence or the proximity of STRE- and Crz1-binding sites, we searched for the CDRE-like sequences [5′-GGGGCT(G/T)-3′] in the promoters that exhibit the bizarre response using the DNA microarray data set produced by Yoshimoto et al. .
First, we looked for the genes whose expression was induced by more than 2-fold when cells were treated with CaCl2 for 30 min and among those, we then looked for genes whose expression was induced further when FK506 is present together with CaCl2. Next, we calculated the ratio of the fold induction between the Ca2+-induced expression, [Ca2+], and the Ca2+ plus FK506-induced expression, [Ca2++FK506]. We set the threshold value for [Ca2++FK506]/[Ca2+] at >1.67, because the value of GLO1 is 1.67 when calculated based on the database. Consequently, 22 genes, including CTT1 (ranking number 4), were found. Finally, we searched for the STRE and CDRE-like sequences in the promoter region (to approx. −500 bp relative to the initiation codon) of the genes. As a result, we found that all of the genes have one or more STREs, whereas only six genes contained the typical CDRE or CDRE-like sequence (Supplementary Table S1 available at http://www.BiochemJ.org/bj/427/bj4270275add.htm). Therefore the co-existence or the proximity of STREs and CDREs seems unnecessary for the bizarre response. We believe that the function of such gene products found in this experiment are not likely to be involved in a certain common biological process; however, some of them are involved in the energy metabolism and we are now investigating the physiological relevance of the bizarre response using these results.
Crz1 destabilizes Msn2/Msn4 in the nucleus in response to Ca2+
A striking feature of the negative effect of Crz1 on Msn2 is that nuclear Crz1 affects the stability of Msn2 and Msn4 in the nucleus when Ca2+ is present (Figure 8A). We call this phenomenon CDMD (Crz1-dependent Msn2/4 degradation). Taking into account the overall results of the present study, we conclude the following mechanism for the CDMD ‘bizarre response’ (Figure 9): Crz1 accumulates in the nucleus upon Ca2+-treatment via calmodulin and calcineurin; in the nucleus it can interact with Msn2/Msn4 and reduce their stability by a yet unknown mechanism. The addition of FK506, inhibiting calcineurin, prevents the simultaneous accumulation of Crz1 and Msn2/Msn4 thereby allowing a further increase in the Ca2+-induced expression of GLO1. Hence, direct disruption of expression of genes encoding calcineurin or Crz1, thereby inhibiting the degradation of Msn2/Msn4, also increases the level of Ca2+-induced expression of GLO1.
Msn2 under conditions of chronic stress or low PKA (protein kinase A) activity is known to be degraded by the 26S proteasome in the nucleus . To examine whether Crz1 enhances the activity of this protein degradation pathway, we treated yeast cells with the proteasome inhibitor MG132. In this experiment, we used an erg6Δ mutant because MG132 take-up is minimal in wild-type cells . As shown in Figure 8(E), Msn2 was degraded in erg6Δ cells following treatment with CaCl2, and this was blocked by FK506. We confirmed that the bizarre response also occurred in the erg6Δ mutant (results not shown). However, the addition of MG132 did not repress the degradation of Msn2 in the presence of CaCl2.
Importantly, the CDMD was not seen in genes where the response to Ca2+ was independent of Hog1, even though the Ca2+-induced expression is still dependent upon Msn2/Msn4 and the STRE (e.g. for GPX1). This means a reduction in the level of Msn2 protein might decrease the integrity of a Hog1–Msn2 complex, which may limit the efficacy of the expression of its STRE-possessing target gene. As far as we could determine, the behaviour of Msn2, with respect to its nuclear localization in response to Ca2+, closely resembled that of Hog1 (Figures 4C and 7C). Hence, the Ca2+-induced expression of GLO1 may be limited by the degradation of Msn2 in addition to the nuclear export of Msn2 and Hog1. The genetic disruption of Crz1, or the inhibition of the nuclear accumulation of Crz1 by inhibiting calcineurin with FK506, block the breakdown of Msn2, thereby stabilizing a Hog1–Msn2 complex to warrant a full response to Ca2+. However, intriguingly, we found that FK506 did not block the degradation of Msn2 in the presence of Ca2+ in Hog1-overexpressing cells (Figure 5D). One clue which explains this result is that we found Msn2 resided in the nucleus for longer periods in Hog1-overexpressing cells in the presence of Ca2+ and FK506; i.e. Crz1 was concentrated in the nucleus following treatment with Ca2+ and redistributed in the cytoplasm after 30–45 min in cells carrying the control vector, but Crz1 resided in the nucleus for 60 min during treatment with CaCl2 in Hog1-overexpressing cells (Supplementary Figure 2B). Durchschlag et al.  have reported that longer residence of Msn2 in the nucleus causes the degradation of Msn2 in a 26S proteasome-dependent manner. Thus in this case we propose that Msn2 might have been degraded in Hog1-overexpressing cells due to the longer retention in the nucleus.
Our findings suggest that an indirect interaction between transcription factors, affecting their stability, may be a regulatory mechanism in the response to stress in addition to other mechanisms (such as the nucleocytoplasmic dynamics, changes to the affinity for their target sequences through post-translational modifications, chromatin remodelling and interactions with RNA polymerase). For example, Williams and Cyert  have reported that the oxidative stress-responsive transcription factor Skn7 regulates the turnover of Crz1; however, the protease involved has not been identified. A network system regulating the stability of transcription factors might exist in yeast cells to warrant an appropriate and distinct stress response to a wide variety of environmental stimuli that partially overlap. This would be the case for a high concentration of Ca2+, which provokes both an osmotic stress response and a Ca2+ signalling response simultaneously. One possible physiological interpretation of the CDMD is that it minimizes unnecessary Hog1–Msn2/Msn4-dependent osmotic responses induced by Ca2+ by destabilizing Msn2/Msn4 through Crz1.
Yoshifumi Takatsume and Takumi Ohdate contributed equally; Yoshifumi Takatsume undertook most of the experimentation shown in Figures 1–7; Takumi Ohdate undertook most of the experimentation shown in Figures 5D, 8 and the Supplementary material and discovered the Crz1-dependent degradation of Msn2/Msn4 in the presence of Ca2+. Kazuhiro Maeta, Wataru Nomura and Shingo Izawa contributed to valuable discussion about the work. Yoshiharu Inoue initiated the study, after discovering the fundamental phenomenon, and wrote the paper.
This research was partially supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan [grant number 20380187].
We thank Dr Y. Kikuchi (Institute for Biomolecular Science, Gakushuin University, Tokyo, Japan) for pRS316+Msn2-GFP, Dr C. Schüller (Department of Biochemistry and Cell Biology, University of Vienna, Vienna, Austria) for pAMG (pAdh1-Msn2-GFP), Dr A. Goldbeter (Faculté des Sciences, Université Libre de Bruxelles, Brussels, Belgium) for pGR247 (pAdh1-Msn4-GFP) and Dr M. Cyert (Department of Biological Sciences, Stanford University, Palo Alto, U.S.A.) for pAMS435 (YEp351+CRZ1) and pAMS463 (GFP–Crz1). We are grateful to Astellas Pharma (formerly Fujisawa Pharmaceutical) for their gift of FK506.
Abbreviations: AP-1, activator protein 1; CDMD, Crz1-dependent Msn2/4 degradation; CDRE, calcineurin-dependent response element; MAPK, mitogen-activated protein kinase; MG, methylglyoxal; SD, synthetic dextrose; STRE, stress-response element; YPD medium, 1% (w/v) yeast extract/2% (w/v) peptone/2% (w/v) glucose medium
- © The Authors Journal compilation © 2010 Biochemical Society